The deployment of artificial intelligence in banking has moved through several distinct phases: from back-office automation to fraud detection, then to chatbots and personalisation engines. The next frontier — autonomous voice agents capable of conducting natural, context-aware conversations with customers — is now operational in Uzbekistan. TBC Bank Uzbekistan, the digital banking arm of the London-listed TBC Bank Group and the leading digital financial ecosystem in Central Asia, has deployed AI agents that make and receive voice calls to handle credit card sales, debt collection, and loan management.
The deployment is one of the most advanced examples of conversational AI in live banking operations anywhere in the emerging market world.
The Technical Foundation Behind the Voice Agents
The technical foundation for this capability was built over several years. TBC Bank Uzbekistan developed proprietary Uzbek-language speech recognition and synthesis models, a finance-specific large language model, and the real-time inference infrastructure needed to run voice interactions at scale. All systems operate on the bank’s own premises — a deliberate architectural choice that ensures both data protection and regulatory compliance.
Unlike institutions that deploy third-party conversational AI platforms adapted from generic international models, TBC Uzbekistan’s voice agents are calibrated specifically for the linguistic and cultural context of the Uzbek market, producing interactions that customers consistently assess as natural and human-like.
The initial application of this technology was in loan repayment reminders, where AI agents began managing early-stage delinquency calls. The results were substantial: by early 2025, these agents were handling over 90% of all early-stage payment reminder interactions, maintaining service quality benchmarks equivalent to human agents while dramatically reducing the cost per interaction.
This track record provided the operational foundation for expanding AI voice deployment into two new domains: credit card sales and proactive credit portfolio management. Both represent significantly more complex conversational tasks than reminder calls, requiring the system to handle objections, answer product questions, and navigate varied customer responses in real time.
From Reminder Calls to Active Sales
The sales application is particularly significant from a commercial standpoint. AI agents are now capable of conducting outbound calls to introduce credit card products, walk customers through relevant features, and progress qualified prospects through the sales funnel — all without human involvement.
Konstantin Kruglov, Head of AI at TBC Bank Uzbekistan, has stated that AI-powered sales chatbots and voice agents deliver ten times the efficiency of human operators. Applied to credit card acquisition, this differential has direct implications for customer acquisition costs, conversion rates, and the speed at which the bank can scale new product lines without expanding headcount proportionally.
Reshaping the Economics of Debt Collection
For debt collection operations, the ability to conduct thousands of personalised, real-time conversations daily — at a fraction of the cost of a human call centre — materially changes the economics of credit portfolio management. Early intervention in delinquency is widely understood to be the most effective stage at which to recover payments; the constraint has traditionally been the cost of making contact at sufficient scale and frequency.
AI voice agents remove that constraint, enabling the bank to reach a far larger proportion of its at-risk borrowers at the earliest possible point in the delinquency cycle. The improvement in collection rates feeds directly into credit loss provisions and, ultimately, into the profitability of the lending book.
Rising Digital Credit Demand in Uzbekistan
The consumer context in which these capabilities are being deployed is important for understanding their full significance. Uzbekistan has a young, digitally active population of over 37 million, increasingly accustomed to managing financial relationships through digital channels. The growing demand for services such as онлайн кредит без залога and online kredit olish tezkor reflects a population that expects credit access to be fast, frictionless, and available on demand without collateral requirements or physical branch visits.
AI voice agents that can engage this population with personalised, immediately responsive interactions are not simply an efficiency tool — they are a product differentiator in a market where consumer expectations are rising faster than many incumbent institutions can adapt.
The broader strategic logic of TBC Bank Uzbekistan’s AI programme extends beyond any individual use case. The bank has built a layered AI infrastructure — speech technology, language models, deployment pipelines — that can be applied progressively across new functions as the technology matures and the organisation’s operational experience accumulates.
Voice agents in sales and collections are today’s application; in-app voice assistants, AI-driven credit decisioning, and automated onboarding are the next wave. Each deployment generates proprietary data that improves the models, creating a feedback loop that compounds the bank’s advantage over competitors who have not made equivalent investments.
Implications for the Regional Banking Landscape
The implications for the wider Central Asian banking market are significant. TBC Bank Uzbekistan’s model demonstrates that AI voice technology is not a distant aspiration for emerging market banks — it is operational today, at scale, and generating measurable commercial returns.
Uzbekistan is a market where digital banking adoption is still accelerating, where incumbent state banks have not yet modernised their customer interaction infrastructure, and where a digitally native institution with proprietary AI capabilities can set the terms of competition for the next decade. TBC Bank Uzbekistan is doing precisely that, and the distance between its AI capabilities and those of its regional peers is growing rather than closing.
The financial performance that accompanies this investment confirms its commercial validity. TBC Bank Uzbekistan’s ecosystem — which includes the digital payments application Payme and the instalment credit platform Payme Nasiya alongside its core banking operations — has grown its user base to over 17 million registered users.
Net profit for the first nine months of 2024 reached $27 million, and the bank is targeting $75 million for the full year of 2025. These figures reflect an institution that is scaling profitably while simultaneously funding one of the most advanced AI programmes in the region — a combination that validates the investment thesis and sets a clear direction for the next phase of growth.
What TBC Bank Uzbekistan is building in Uzbekistan is not simply a more efficient version of conventional banking. It is a blueprint for how financial institutions in high-growth emerging markets can use proprietary AI infrastructure to compress the development timeline, outpace incumbents, and establish competitive positions that are structurally difficult to challenge.
Voice AI is the current leading edge of that effort — but the architecture being assembled beneath it will support capabilities that extend well beyond what is visible today. The institutions paying close attention to what is happening in Uzbekistan’s banking market will find more there than they expect.
